Empowering Image Recovery_ A Multi-Attention Approach
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929307154120704 |
|---|---|
| author | Wen, Juan Li, Yawei Zhang, Chao Hou, Weiyan Timofte, Radu Van Gool, Luc |
| author_facet | Wen, Juan Li, Yawei Zhang, Chao Hou, Weiyan Timofte, Radu Van Gool, Luc |
| contents | We propose Diverse Restormer (DART), a novel image restoration method that effectively integrates information from various sources (long sequences, local and global regions, feature dimensions, and positional dimensions) to address restoration challenges. While Transformer models have demonstrated excellent performance in image restoration due to their self-attention mechanism, they face limitations in complex scenarios. Leveraging recent advancements in Transformers and various attention mechanisms, our method utilizes customized attention mechanisms to enhance overall performance. DART, our novel network architecture, employs windowed attention to mimic the selective focusing mechanism of human eyes. By dynamically adjusting receptive fields, it optimally captures the fundamental features crucial for image resolution reconstruction. Efficiency and performance balance are achieved through the LongIR attention mechanism for long sequence image restoration. Integration of attention mechanisms across feature and positional dimensions further enhances the recovery of fine details. Evaluation across five restoration tasks consistently positions DART at the forefront. Upon acceptance, we commit to providing publicly accessible code and models to ensure reproducibility and facilitate further research. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_04617 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Empowering Image Recovery_ A Multi-Attention Approach Wen, Juan Li, Yawei Zhang, Chao Hou, Weiyan Timofte, Radu Van Gool, Luc Computer Vision and Pattern Recognition 68T07 (Primary) 168T45 (Secondary) I.4.4 We propose Diverse Restormer (DART), a novel image restoration method that effectively integrates information from various sources (long sequences, local and global regions, feature dimensions, and positional dimensions) to address restoration challenges. While Transformer models have demonstrated excellent performance in image restoration due to their self-attention mechanism, they face limitations in complex scenarios. Leveraging recent advancements in Transformers and various attention mechanisms, our method utilizes customized attention mechanisms to enhance overall performance. DART, our novel network architecture, employs windowed attention to mimic the selective focusing mechanism of human eyes. By dynamically adjusting receptive fields, it optimally captures the fundamental features crucial for image resolution reconstruction. Efficiency and performance balance are achieved through the LongIR attention mechanism for long sequence image restoration. Integration of attention mechanisms across feature and positional dimensions further enhances the recovery of fine details. Evaluation across five restoration tasks consistently positions DART at the forefront. Upon acceptance, we commit to providing publicly accessible code and models to ensure reproducibility and facilitate further research. |
| title | Empowering Image Recovery_ A Multi-Attention Approach |
| topic | Computer Vision and Pattern Recognition 68T07 (Primary) 168T45 (Secondary) I.4.4 |
| url | https://arxiv.org/abs/2404.04617 |